Goals

Firstly, I need to load packages I am going to use for this assignment.

library(gapminder)
library(tidyverse)
library(knitr)
library(plotly)
library(gridExtra)

Part 1: factor management

Drop Oceania

Before Drop Oceania

Before Dropp Oceania, let’s look at levels of continent variable and see how many entries correspond to each continent.

p1<-gapminder$continent 
levels(p1) #levels of continent variable
## [1] "Africa"   "Americas" "Asia"     "Europe"   "Oceania"
kable(fct_count(p1),col.names = c('continent','number of entry')) #number of entries for each continent
continent number of entry
Africa 624
Americas 300
Asia 396
Europe 360
Oceania 24 
a<-nlevels(p1) #number of levels
b<-nrow(gapminder) #number of entries after removing oceania
kable(data.frame(number_of_levels = a, number_of_entries = b))
number_of_levels number_of_entries
5 1704

Drop Oceania

We need to remove observations associated with the continent of Oceania. Additionally, remove unused factor levels.

p2<-gapminder%>%
  filter(continent != 'Oceania')%>%
  droplevels()

After Drop Oceania

Now we can look at the level of continent variable and see how many entries correspond to each continent.

p3<-p2$continent
levels(p3) #levels of continent variable
## [1] "Africa"   "Americas" "Asia"     "Europe"
kable(fct_count(p3),col.names = c('continent','number of entry')) #number of entries for each continent
continent number of entry
Africa 624
Americas 300
Asia 396
Europe 360 
a<-nlevels(p3) #number of levels
b<-nrow(p2) #number of entries after removing oceania
kable(data.frame(number_of_levels = a, number_of_entries = b))
number_of_levels number_of_entries
4 1680

As expected, the number of rows has dropped by 24, corresponding to the number of observations for Oceania. The number of levels for continent has dropped by 1(Oceania level is dropped).

Reorder the levels of conitnent

Use the forcats package to change the order of the factor levels, based on a principled summary of one of the quantitative variables. Consider experimenting with a summary statistic beyond the most basic choice of the median.

In this question, I will reorder the data based on gdp range, which is difference between maximum gdpPercap and minimum gdpPercap of each continent.

Firstly, I need to use group_by() and summarize() functions to calculate gdp range for each continent.

p4<-group_by(gapminder,continent)%>%
  summarize(gdp_range = max(gdpPercap)-min(gdpPercap))
levels(p4$continent)
## [1] "Africa"   "Americas" "Asia"     "Europe"   "Oceania"
kable(p4)
continent gdp_range
Africa 21710.05
Americas 41750.02
Asia 113192.13
Europe 48383.66
Oceania 24395.77 
ggplot(p4,aes(x=gdp_range,y=continent))+
  geom_point(color='red')

Now I want to use fct_reorder() function to reorder the factor levels of each continent by their gdp_range variable.

p5<-mutate(p4,continent=fct_reorder(continent,gdp_range))
levels(p5$continent)
## [1] "Africa"   "Oceania"  "Americas" "Europe"   "Asia"
kable(p5)
continent gdp_range
Africa 21710.05
Americas 41750.02
Asia 113192.13
Europe 48383.66
Oceania 24395.77 
ggplot(p5,aes(x=gdp_range,y=continent))+
  geom_point(color='red')

According to table, we can find fct_reorder() function reorder factor levels of continent and the plot is ordered in the desired way, but it does not rearrange data in the dataframe.

Next, I will use arrange() function to arrange the order of data by their gdp_range variable.

p6<-p4%>%
  arrange(gdp_range)
levels(p6$continent)
## [1] "Africa"   "Americas" "Asia"     "Europe"   "Oceania"
kable(p6)
continent gdp_range
Africa 21710.05
Oceania 24395.77
Americas 41750.02
Europe 48383.66
Asia 113192.13 
ggplot(p6,aes(x=gdp_range,y=continent))+
  geom_point(color='red')

Obviously, arrange() function arranges the data in an increading order of gdp_range variable, but it does not change the factor levels of contiennt.

Now we can combine arrange() and fct_reorder to see what happens.

p7<-p4%>%
  arrange(gdp_range)%>%
  mutate(continent=fct_reorder(continent,gdp_range))
levels(p7$continent)
## [1] "Africa"   "Oceania"  "Americas" "Europe"   "Asia"
kable(p7)
continent gdp_range
Africa 21710.05
Oceania 24395.77
Americas 41750.02
Europe 48383.66
Asia 113192.13 
ggplot(p7,aes(x=gdp_range,y=continent))+
  geom_point(color='red')

We can find that both table and plot are ordered in the desired way.

Conclusion:

  • fct_reorder() function changes factor levels of continent but it does not rearrange data in the dataframe. It has an effect on how variables are organized on plots.
  • arrange() function arranges the data in an increading order of gdp_range variable, but it does not change the factor levels of contiennt. So the how variables are organized on plots are not affected by arrange() function.
  • combining fct_reorder() and arrange() can both changes factor levels of continent and arranges the data in an increading order of gdp_range variable.

Part 2: File I/O

Explore write_csv() and read_csv function

I’m going to try writing the re-ordered dataframe to a csv file and then reload it.

re_order <- group_by(gapminder,continent)%>%
            summarize(gdp_range = max(gdpPercap)-min(gdpPercap))%>%
            mutate(continent=fct_reorder(continent,gdp_range))
str(re_order) # look at the structure of re-ordered dataframe
## Classes 'tbl_df', 'tbl' and 'data.frame':    5 obs. of  2 variables:
##  $ continent: Factor w/ 5 levels "Africa","Oceania",..: 1 3 5 4 2
##  $ gdp_range: num  21710 41750 113192 48384 24396
levels(re_order$continent) # look at the levels of continent variable
## [1] "Africa"   "Oceania"  "Americas" "Europe"   "Asia"

Write and reload .csv files.

write_csv(re_order,"a1.csv") 
data<-read_csv("a1.csv")
## Parsed with column specification:
## cols(
##   continent = col_character(),
##   gdp_range = col_double()
## )

Look at the structure of reloaded dataframe.

str(data)
## Classes 'tbl_df', 'tbl' and 'data.frame':    5 obs. of  2 variables:
##  $ continent: chr  "Africa" "Americas" "Asia" "Europe" ...
##  $ gdp_range: num  21710 41750 113192 48384 24396
##  - attr(*, "spec")=List of 2
##   ..$ cols   :List of 2
##   .. ..$ continent: list()
##   .. .. ..- attr(*, "class")= chr  "collector_character" "collector"
##   .. ..$ gdp_range: list()
##   .. .. ..- attr(*, "class")= chr  "collector_double" "collector"
##   ..$ default: list()
##   .. ..- attr(*, "class")= chr  "collector_guess" "collector"
##   ..- attr(*, "class")= chr "col_spec"

Obviously, the continent variable is not a factor but a list. So write_csv()\read_csv() changes the the attributes of factor variable in dataframe.

Explore saveRDS() and readRDS()

Now let’s save re-ordered dataframe to a file and reopen it again, this time using saveRDS()/readRDS().

str(re_order)
## Classes 'tbl_df', 'tbl' and 'data.frame':    5 obs. of  2 variables:
##  $ continent: Factor w/ 5 levels "Africa","Oceania",..: 1 3 5 4 2
##  $ gdp_range: num  21710 41750 113192 48384 24396
levels(re_order$continent)
## [1] "Africa"   "Oceania"  "Americas" "Europe"   "Asia"

Write and reload .rds files.

saveRDS(re_order,"a2.rds")
data<-readRDS("a2.rds")

Look at the structure of reloaded dataframe.

str(data)
## Classes 'tbl_df', 'tbl' and 'data.frame':    5 obs. of  2 variables:
##  $ continent: Factor w/ 5 levels "Africa","Oceania",..: 1 3 5 4 2
##  $ gdp_range: num  21710 41750 113192 48384 24396
levels(data$continent)
## [1] "Africa"   "Oceania"  "Americas" "Europe"   "Asia"

The dataframes are completely identical. Functions saveRDS()/readRDS() preserved the order of the factor continent, unlike write_csv()\read_csv().

Explore dput() and dget()

Look at the structure of re-ordered dataframe and levels of continent variables.

str(re_order)
## Classes 'tbl_df', 'tbl' and 'data.frame':    5 obs. of  2 variables:
##  $ continent: Factor w/ 5 levels "Africa","Oceania",..: 1 3 5 4 2
##  $ gdp_range: num  21710 41750 113192 48384 24396
levels(re_order$continent)
## [1] "Africa"   "Oceania"  "Americas" "Europe"   "Asia"

Write and reload .R files.

dput(re_order,"a3.R")
data<-dget("a3.R")

Look at the structure of reloaded dataframe.

str(data)
## Classes 'tbl_df', 'tbl' and 'data.frame':    5 obs. of  2 variables:
##  $ continent: Factor w/ 5 levels "Africa","Oceania",..: 1 3 5 4 2
##  $ gdp_range: num  21710 41750 113192 48384 24396
levels(data$continent)
## [1] "Africa"   "Oceania"  "Americas" "Europe"   "Asia"

We can see that dataframe before saving is the same as that after loading, so dput()\dget() does not destroy any information in dataframe.

Part 3: Visualization design

In part I, I use scatter plots to show difference of arrange() and fct_reorder function. In this part, I will use bar plots to visulize gdp range (difference between maximum gdp per capita and minimum gdp per capita) in different continents.

# group and summarize gdp range data in different continents
p8 <- group_by(gapminder,continent)%>%
      summarize(gdp_range = max(gdpPercap)-min(gdpPercap))

# bar plot for original data
pl1<-ggplot(p8,aes(x=continent,y=gdp_range))+
     geom_bar(aes(fill=continent),stat="identity",position="dodge")+
     theme(plot.title = element_text(size=14,hjust=0.5),
           axis.text.x = element_text(angle = 45, hjust = 1))+
     labs(x="continent",
          y="gdp_range",
          title="before process")

# arrange gap_range in an increasing order
p9 <- arrange(p8,gdp_range)
# bar plot for arranged data
pl2<-ggplot(p9,aes(x=continent,y=gdp_range))+
     geom_bar(aes(fill=continent),stat="identity",position="dodge")+
     theme(plot.title = element_text(size=14,hjust=0.5),
           axis.text.x = element_text(angle = 45, hjust = 1))+
     labs(x="continent",
          y="gdp_range",
          title="arrange")

# re-order factor levels of continent variable
p10 <- mutate(p8,continent = fct_reorder(continent, gdp_range))
# bar plot for re-ordered data
pl3<-ggplot(p10,aes(x=continent,y=gdp_range))+
     geom_bar(aes(fill=continent),stat="identity",position="dodge")+
     theme(plot.title = element_text(size=14,hjust=0.5),
           axis.text.x = element_text(angle = 45, hjust = 1))+
     labs(x="continent",
          y="gdp_range",
          title="reorder factor levels")

# combine arrange and re-order 
p11<-arrange(p8,gdp_range)%>%
     mutate(continent = fct_reorder(continent, gdp_range))
# bar plot for arrange and fct_reorder
pl4<-ggplot(p11,aes(x=continent,y=gdp_range))+
     geom_bar(aes(fill=continent),stat="identity",position="dodge")+
     theme(plot.title = element_text(size=14,hjust=0.5),
           axis.text.x = element_text(angle = 45, hjust = 1))+
     labs(x="continent",
          y="gdp_range",
          title="arrange and reorder")

# show 4 plots in one page
grid.arrange(pl1,pl2,pl3,pl4,nrow=2)

Next, I will compare the difference of ggplot2 graph and plotly graph.

First, I will make a scatterplot to compare life expectanct of China, Canada and Germany using ggplot2.

p12<-filter(gapminder, country %in% c('China','Germany','Canada'))%>%
  select(year,country,lifeExp)
(pl5<-ggplot(p12,aes(x=year,y=lifeExp,color=country))+
      geom_point()+
      geom_line()+
      scale_x_continuous(limits=c(1952,2007),breaks=seq(1952,2007,5))+
      theme(axis.text.x = element_text(angle = 45, hjust = 1),
           plot.title = element_text(size=14,hjust=0.5))+
      labs(x="country",
           y="life expectancy",
           title="Life expectancy of China, Canada, Germany from 1952 to 2007"))

Now I will convert it into plotly graph.

ggplotly(pl5)

Also, I want to make 3D plot of above plot using plot_ly().

plot_ly(p12,
        x=~year,
        y=~lifeExp,
        z=~country,
        type='scatter3d',
        mode='markers')

plotly is amazing! It allows people to quickly create beautiful, reactive D3 plots that are particularly powerful in websites and dashboards. we can also hover our mouse over the plots and see the data values, zoom in and out of specific regions, and capture stills.

Part 4: Writing figures to file

Firstly, I will use density plot to visualize the distribution of GDP per capita for each continent from year 1952 to 2007 and then save it as an PNG image.

# plot1
plot1<-ggplot(gapminder,aes(x=gdpPercap))+
  geom_density(color='red')+
  facet_wrap(~continent)+
  scale_x_continuous(breaks = seq(0, 120000, 20000),
                     labels = as.character(seq(0, 120000, 20000)),
                     limits = c(0,120000))+
  theme(axis.text.x = element_text(angle = 45, hjust = 1),
        plot.title = element_text(size=14,hjust=0.5))+
  labs(x="GDP per capita",
       y="density",
       title="Distribution of GDP per capita for each continent from 1952 to 2007")
# save plot1 as an PNG image
ggsave("plot1.png",
       plot=plot1,
       width = 10, 
       height = 7,
       scale = 1.2)

Then I will load and embed it in my report.

plot1

plot1

Secondly, I will use histogram plot to visualize the distribution of GDP per capita for each continent and then save this as a PDF image.

# plot2
plot2 <- filter(gapminder,year %in% c(1952,2007))%>%
  ggplot(aes(x=gdpPercap))+
  geom_histogram(fill='blue',alpha=0.3,bins=45)+
  facet_grid(continent~year)+
  scale_x_continuous(breaks = seq(0, 60000, 10000),
                     labels = as.character(seq(0, 60000, 10000)),
                     limits = c(0,60000))+
  theme(axis.text.x = element_text(angle = 45, hjust = 1),
        plot.title = element_text(size=14,hjust=0.5))+
  labs(x="GDP per capita",
       y="number of countries",
       title="Distribution of GDP per capita for each continent in 1952 and 2007")
# save plot2 as a pdf image
ggsave("plot2.pdf",
       plot=plot2,
       width = 10, 
       height = 7,
       scale = 1.2)
## Warning: Removed 1 rows containing non-finite values (stat_bin).

Then I will load and embed it in my report.

plot2